Rasmussen Tracking Poll [Obama vs McCain]

[Gustaf]

That's a very sudden shift in their favorability numbers (for both candidates). Odd.

My thoughts exactly. Very odd. Interesting that Obama seems to be doing so poorly with Hispanics. The white number, on the other hand, isn't so bad for a Democrat is it?

[Gustaf]

That's a very sudden shift in their favorability numbers (for both candidates). Odd.

My thoughts exactly. Very odd. Interesting that Obama seems to be doing so poorly with Hispanics. The white number, on the other hand, isn't so bad for a Democrat is it?

If Kerry had those breakdowns in 2004, he would have won Ohio and the election (assuming a uniform swing of demographics), as well as possibly other states. Kerry did lose whites by IIRC eighteen points in 2004. Bear in mind of course that the sample sizes for blacks and Hispanics both are quite small, only a couple hundred at most.

Yes, yes, of course. Still, given undecideds and MoE the difference between 13 and 18 isn't enormous. Besides, I suspect Obama whites will be spread unevenly across the states compared to Kerry's...

[Gustaf]

Wtf happened with Obama's favourables. That jump is a bit insane in such a short period of time.

[Gustaf]

Using fake numbers and arbitrary assumptions, Alcon gets the results he wants.

While doing your abracadra routine, you still have NOT dealt with the problem of allocating the undecideds.

You are merely assuming that they will vote the same way that the decideds are voting, weighted for one demographic!

If you delve into (admittedly delayed) information from the University of Michigan Survey Research Center (they've been at it for sixty years), you will find that undecideds are a completely different type than the decideds.

Carl,

since you misunderstood what Alcon said you could still save some face by acknowledging that. The undecideds in this case are not undecideds in the presidential race but undecideds in the senatorial race. Those undecideds may have a preference in the presidential race. Alcon forgot to include those in his calculation. The term extrapolation was a mistake on his part (as he noted) since that is not what he did. He merely used the subsamples presented by Rasmussen Reports (which add up to the total sample) to calculate the result for the whole sample. This is commonly done by most of us whenever a poll releases results from a set of subsamples that add up to the whole sample before they release the topline numbers. There is nothing statistically unsound here, since there are no new assumptions introduced.

[Gustaf]

Using fake numbers and arbitrary assumptions, Alcon gets the results he wants.

While doing your abracadra routine, you still have NOT dealt with the problem of allocating the undecideds.

You are merely assuming that they will vote the same way that the decideds are voting, weighted for one demographic!

If you delve into (admittedly delayed) information from the University of Michigan Survey Research Center (they've been at it for sixty years), you will find that undecideds are a completely different type than the decideds.

Carl,

since you misunderstood what Alcon said you could still save some face by acknowledging that. The undecideds in this case are not undecideds in the presidential race but undecideds in the senatorial race. Those undecideds may have a preference in the presidential race. Alcon forgot to include those in his calculation. The term extrapolation was a mistake on his part (as he noted) since that is not what he did. He merely used the subsamples presented by Rasmussen Reports (which add up to the total sample) to calculate the result for the whole sample. This is commonly done by most of us whenever a poll releases results from a set of subsamples that add up to the whole sample before they release the topline numbers. There is nothing statistically unsound here, since there are no new assumptions introduced.

First, I realize that you and Alcon does some funny things with numbers to get the results you want, but that does not make such practices valid or sound.

What Alcon did was to assume that if adjusted for one demographic factor (gender) he could project the undecided vote based on the decided vote.

No, that was not what Alcon did. I have bolded the part in my post where I explain what he did. Let's say you have a poll which says how Texas whites are going to vote and how Texas blacks are going to vote and how Texas Hispanics are going to vote. If we assume that there are no other ethnic groups in Texas we could then calculate the overall standing between the two candidates based on their standings within these subsamples. If one did this but forgot to include Hispanics, looking only at blacks and whites, the result would however be wrong. And that is what Alcon did in this case.

Now, I would like to remind you that I have never used name-calling or gone out of my way to attack you and I would prefer that you didn't insinuate things such as me doing "funny things" with numbers. I don't and neither did Alcon in this case.

[Gustaf]

Well, congratulations Alcon. I think your point actually ended up getting through.

[Gustaf]

I just read in USA Today this morning that Obama expects to have a shot in South Carolina and Georgia - "and that they are looking at Tennessee and Alabama".

If this is all true, and the blue states are the blowouts that the polls are telling us they are, that begs the question - how is McCain tied or within the margin of error in the national tracking numbers in both Gallup and Rasmussen?  Especially when you throw in the supposed closeness of Indiana, Montana, the Dakotas and Obama double digit leads in purple states like Iowa and Wisconsin.  Add in a healthy Quinnipiac lead for Obama in the populous state of Florida.

I'm just saying it doesn't add up.  Someone help me here.

Is McCain up 30 in Texas?  Nope.  I'm hearing just 7 or 8 according to Rasmussen and not the 23 Bush beat Kerry by.

you are correct.  It is impossible that every poll conducted this election season is ultimately the exact margin that will occur in the fall.

It is also impossible that all of the polls most favorable to obama can be true in the states, while the tracking polls are also accurate.

Most likely the answers vary from maybe obama's up more than the tracking polls say to maybe some of the red states aren't as close as they are currently polling to maybe some of the swing states are still close even though polls show obama winning handily to some of the blue states obama will likely carry are closer than they are polling or closer than they were last time around... or some combination of the above. 

Maybe a reasonable exercise would be to calculate margins in all 50 states with projected turnout and see where that leads in terms of national vote generally.  I personally don't have the wherewithall or time to do it.  Maybe you could try it yourself.

I've been doing that for a month. The numbers add up perfectly. Or at least they did till the tracking polls got close recently.

The state polls currently show Obama ahead by 4.37% nationally. Till recently that was his margin in both Gallup and Rasmussen nationally. It could be that state polls suck and that the swing is so recent that the small number of state polls we have seen since then hasn't been enough to shift the national average yet.

[Gustaf]

Wednesday, August 20, 2008:

General: Head-to-Head

Obama: 45% / 47%, including leaners (nc, nc)
McCain: 42% / 46%, including leaners (nc, +1)

Favorability

Obama: 56% favorable, 43% unfavorable (+2, -2)
McCain: 55% favorable, 43% unfavorable (nc, nc)

New data will be released for the New Hampshire Senate race at 3:00 p.m. Eastern today. Presidential race results for Maryland and New Hampshire will be released at 5:00 p.m. Eastern.

First Maryland poll in ages, yippee!

Isn't it the first Maryland poll ever (ever meaning since the idea of Obama v McCain surfaced...) from Rasmussen?

[Gustaf]

He's talking about the independent vote, not the overall vote. Given the current voter registration numbers he is almost certainly correct (even though McCain's Rep majority will probably be higher than Obama's Dem majority).

In an overall tie I'd give it to McCain in the EC, because I think he has a small edge there relative to the national popular vote. But a complete tie is unlikely, once everything is factored in.

[Gustaf]

Like Torie says, it's generally better to use resources now than later, contrary to what many think. Of course, this presumes general growth of the economy, something which isn't necessarily true if we run out of energy.

[Gustaf]

Can someone remind me which presidential nominee has been aligned with the very personification of ineptitude (Mr 34%, according to Rasmussen) 90% of the time in the Senate? Is contributing to the mess really worthy of a 22% net approval rating?

Four more years of America's international standing in the gutter; four more years of relentless attacks on middle class living standards; four more years of blue collar jobs being outsourced. Ugh

I'm going to wipe the floor with McCain if he wins - and things aren't a hell of a lot better in four years time. And that's a promise

Dave

What are you gonna do? Torture him?

[Gustaf]

so conservative democrats are saying that they will vote or answer for McCain for the moment?

The point is rather that they may identify themselves as independents instead of Democrats at the moment, while conservative independents may to a greater extent be identifying as Republicans now.

[Gustaf]

Rasmussen has been hugely stable.  It doesn't swing a lot.  There is no midweek/weekend bounce. 

Right now, it is a tie.

Mensa are so lucky to have you as a memeber.

Yes, and very fortunately, you cannot get it.

You mean, he "cannot get in"?

"it" could theoretically refer to Mensa membership. But you're not really in a position to call out people on making sense in their posting.

[Gustaf]

Rasmussen has been hugely stable.  It doesn't swing a lot.  There is no midweek/weekend bounce. 

Right now, it is a tie.

Mensa are so lucky to have you as a memeber.

Yes, and very fortunately, you cannot get it.

You mean, he "cannot get in"?

No, double meaning.

Please, Gustav, he's helping to illustrate a point. 

You're a retard.

There's only one meaning in that.

Look, "Iosif", I don't know who you are but that's not a very constructive attitude. If you have a disagreement with JJ or want to criticize something he said, do it. But don't just call someone a retard.

[Gustaf]

Obama should just give up allready. No fancy statistical mumbo-jumbo can change the FACT that he's done.

[Gustaf]

@Hawk: Don't preach to me. He is a clear cut socialist.

Bullsh**t. I tell it like it is and if you don't like it. Tough

You don't "tell it like it is" you voice your opinion as if it were an undisputable fact, which is a bit conceited.

Anyway, I doubt Obama is an idelogical socialist. I don't think he has any kind of inner restraint as to what the size or influence of the government should be either though.

But enough hijacking people. I want to be able to read the polls.

[Gustaf]

Yep. And Gallup, once again, sucked completely. That firm seems to have really, really lost it.

[Gustaf]

Well this makes ChrisNJ look foolish, which is a surprise to no one.

Rasmussen's state polls were terrible and had a McCain lean. I was right.

They were not terrible and the supposed McCain lean was one of the smallest biases of all pollsters and obviously just random noise.

But how did Florida and West Virginia compare to national average again?

[Gustaf]

I don't really care that Rasmussen underpolled Obama in Hawaii or Massachusetts or Alabama or something. Those were going to be blow-outs and no one cared. If you look at the swing states, then Rasmussen had a small, but consistent Republican bias of about 2-4 points. Same with Mason-Dixon, who really only did well in Florida (which, to their credit, they got exactly).

Eh...so your argument is what exactly? Rasmussen on purpose tried to inflate McCain's poll numbers in swing states so that they could look incompetent and then tried to cover it up by inflating Obama's poll numbers in other states so as to make them look even worse?

[Gustaf]

I don't really care that Rasmussen underpolled Obama in Hawaii or Massachusetts or Alabama or something. Those were going to be blow-outs and no one cared. If you look at the swing states, then Rasmussen had a small, but consistent Republican bias of about 2-4 points. Same with Mason-Dixon, who really only did well in Florida (which, to their credit, they got exactly).

Eh...so your argument is what exactly? Rasmussen on purpose tried to inflate McCain's poll numbers in swing states so that they could look incompetent and then tried to cover it up by inflating Obama's poll numbers in other states so as to make them look even worse?

Er... no? When did I say anything like that? I don't believe that Rasmussen has some secret pro-Republican agenda and that he's fixing his polls. He just didn't do as well this cycle as he has in past cycles, and something about how he took his polls gave them a bit of a Republican lean.

I looked at the final margin in Rass's polls versus the final real margin in 18 swing states (CO, NV, NM, AZ, MT, MN, IA, MO, WI, IN, MI, OH, PA, VA, NC, FL, and NH). In only four of these states was Obama's margin over-polled, AZ, GA, MO, and MN (and in GA and MO, the difference was less than 1, which could have just been a matter of rounding). In all the other states he under-polled Obama, sometimes by small margins of 1-2, but more than half the time by larger margins of 3-8.

Yeah...and I looked at all of the states and I didn't get any Republican bias.

But then I assume that you are arguing that Rasmussen used different methodologies in close states as opposed to non-close states? (and close states here mean states he may have thought were close, and not states that actually ended up close).

[Gustaf]

States that neither candidate campaigns in are more susceptible to large, random swings, so I don't think looking at those really tells us anything. The swing states, where most people have made up their mind by election day, should be a lot easier to poll, and I think that how a pollster does in those states is much more important.

I'm not sure I agree. If anything, the opinion in those states should be pretty stable since it isn't being hammered by the campaigning? Do you have any evidence for this claim? It seems to me like you're awfully close to arbitrarily throwing out data that doesn't fit your conclusions.

[Gustaf]

How is there a "Presidential Approval rating" for Obama when he isn't even President...

I can only suppose it means they approve of how he's handling the transition

I believe it refers to "President of the Universe", a title that is Obama's for life.